Difference between revisions of "Team:Marburg/Design"

Line 230: Line 230:
 
   <div class="content">
 
   <div class="content">
 
<div class="collapsible">
 
<div class="collapsible">
<div class="btn_expand">Accelerated metabolic Engineering</div>
+
<div class="btn_expand">Our vision of Accelerating Metabolic Engineering</div>
 
<div class="content">
 
<div class="content">
  
Line 427: Line 427:
  
 
 
<div class="skipTarget" skipname="Biosensors"></div>
+
<div class="skipTarget" skipname="Biosensors for rapid product screening"></div>
 
<div class="collapsible">
 
<div class="collapsible">
   <div class="btn_expand">Biosensors</div>
+
   <div class="btn_expand">Biosensors for rapid product screening</div>
 
   <div class="content">
 
   <div class="content">
  

Revision as of 17:46, 14 October 2018

Design

What does creating task specific lab strains, building a novel golden-gate based toolbox and envisioning a synthetic metabolic pathway have in common?
All these approaches require intense theoretical groundwork. Many hours have to be spent with screening literature, summarizing the current state of knowledge and finally, developing own ideas. Constructs have to be designed and experiments have to be planned.
All this preliminary work has to be completed before the lab work can be truly started. In addition to our initial theoretical work, we routinely iterated the steps: designing, building, testing and learning throughout the whole project.

On this page, we want to provide you the theoretical groundwork that we did to design our projects. We want to demonstrate that we followed Synbio design principles and conceived our project based on the current state of knowledge.

Design of the Marburg Collection

Overview of the Marburg Collection

Overview of the Marburg Collection

A major goal of our project was to create a collection of characterized parts for use in V. natriegens to provide other iGEM teams, as well as the whole scientific community a toolbox for the rational design of metabolic pathways, genetic circuits or any other DNA construct.
Our toolbox was initially conceived for being used in V. natriegens. However, while designing the Marburg Collection, we realized that a toolbox with maximum flexibility can easily be used in more than one bacterial species. Because alternative bacterial chassis, apart from E. coli, are gaining increasing importance (Kim et al. 2016), a toolbox that is compatible with more than one bacterial species enables scientists to work with organisms that have the exact properties needed for specific applications. Our vision is to establish the Marburg Collection as the first broad-host range golden-gate-based cloning toolbox.
Figure 1: LVL1 assembly using the Marburg Collection
De novo assembly of LVL1 plasmids from eight basic parts.
We investigated existing bacterial toolboxes like EcoFlex (Moore et al. 2016), CIDAR (Iverson et al. 2016) and iGEMs PhytoBrick system, which is closely related to the GoldenBraid cloning (Sarrion-Perdigones et al. 2011). All these toolboxes are based on predefined vectors that preset the origin of replication and the antibiotic resistance cassette thus preventing their use in organisms that are not compatible with either feature. We wanted to give scientists the freedom of choice to select oris and resistance cassettes that suit their organisms and applications best. This flexibility is possible because our toolbox does not rely on classical “backbones” but instead functions by the complete de novo assembly of at least eight basic genetic parts.

All basic parts like promoters or resistance parts are stored in LVL0 plasmids. The assembly of a plasmid comprising a single transcription unit is done by assembling at least eight parts resulting in one LVL1 plasmid. One to five LVL1 plasmids can then be used for a subsequent round of assembly to obtain a multigene LVL2 plasmid which could already harbor a full synthetic metabolic pathway consisting of up to five enzymes. Our toolbox even allows more rounds of assembly, each combining up to five constructs of the previous level. This enormous cloning capacity results theoretically in an infinite number of transcription units that can be assembled in a small amount of time.
Figure 2: Hierarchical cloning is facilitated by subsequent Golden Gate reactions.
Basic building blocks like promoters or terminators are stored in level 0 plasmids. Parts from each category of our collection can be chosen to built level 1 plasmids harboring a single transcription unit. Up to five transcription units can be assembled into a level 2 plasmid.
An additional layer of flexibility in the construction of LVL2 plasmids is added to our toolbox by our 5’- and 3’- Connectors which flank the transcription units in LVL1 plasmids. These connectors provide the fusion sites for the assembly of LVL2 plasmids. By selecting the correct connectors, the user can define the order as well as the orientation of each transcription unit. The ability to assign an orientation to a transcription unit is supposed to reduce the influence of neighboring unit on each other caused by transcriptional read through and DNA supercoiling. Moreover, our connectors are designed to function as genetic insulators. They consist of 300 bp “neutral DNA” flanked by strong transcriptional terminators to separate each transcription unit in a LVL2 plasmid. We believe that this novel feature is an important step towards building genetic constructs in a rational and predictable manner.

Construction of novel part entry vectors

Construction of novel part entry vectors

All LVL0 parts have to be stored in plasmids to allow for amplification and long term storage. To create new LVL0 parts, a PCR product or annealed oligos are cloned into a part entry vector. This vector harbours the resistance and ori that are required for selection and propagation. Furthermore, part entry vectors can be designed in a way that they contain a dropout. This dropout can be a transcription unit for a marker that generates a visible output. The first golden-gate-based toolbox MoClo (Weber et al. 2011) used a LacZ alpha transcription unit which can be used for blue white screening in many E. coli cloning strains. This concept was also adapted by iGEMs PhytoBrick system. During the cloning of LVL0 parts, this dropout is replaced by the desired part. When the cloning reaction is transformed into a suitable E. coli strain and the cells are plated on agar plates with supplemented IPTG and X-Gal. Colonies transformed with the religated entry plasmid appear blue while white colonies most probably contain the correctly assembled plasmid. The LVL0 part entry vector in iGEMs PhytoBrick system (BBa_P10500) has been designed as described and can be used for blue white screening.
We appreciate the approach of using part entry plasmids with dropouts but, for two reasons, we think that LacZ is not an optimal reporter. First, blue white screening requires the two expensive chemicals IPTG and X-Gal which have to be added to the agar plates. Second, blue white screening is restricted to E.coli strains with an incomplete lac operon that is complemented by the LacZ alpha fragment that is expressed from the plasmid (Langley et al. 1975). Consequently blue white screening is not compatible with a V. natriegens wild type strain (Link zu Improvement Page).
Figure 2: Plasmid map of K2560002
This part enables creation of new LVL0 parts by replacing the parts sequence with the desired insert. The sfGFP expression cassette is flanked by BsmBI recognition sites
We overcame both drawbacks by replacing the LacZ alpha dropout of BBa_P10500 by a RFP and sfGFP dropout and thus created the two parts BBa_K2560001 and BBa_K2560002, respectively. Like BBa_P10500, our two new dropout parts are located in a derivative of pSB1C3 which is extended with two BsaI recognition sites. The dropout parts are flanked by BsmBI recognition sites that are required for replacing the dropout by the desired LVL0 part. The resulting LVL0 parts are classified as PhytoBricks and are RFC10 and RFC25 compatible, assuming that the part itself does not contain prohibited recognition sites. The resulting plasmids can be used to clone and store all LVL0 parts. However, for resistance parts a different approach enables more convenient LVL1 cloning. In the LVL1 golden-gate-reaction, at least eight parts are combined in a single tube together with the required enzymes. All parts, except the resistance part, are stored in the previously described pSB1C3 derivate which confers a chloramphenicol resistance. Obviously, the plasmid that provides the new resistance cassette for the LVL1 plasmid (e.g. a Cas9 plasmid with kanamycin resistance), has to contain the same resistance cassette (kanamyicin in this example). Because cloning is never 100 % efficient, re-ligation of LVL0 plasmids is a common event, and in case of the LVL0 resistance part, results in false positive colonies which do not contain the desired LVL1 plasmid.
We developed a solution for this problem by creating the novel resistance entry vectors BBa_K2560005 and BBa_K2560006. These plasmids contain a RFP and sfGFP dropout, respectively, and a chloramphenicol resistance cassette that is flanked by BsaI and BsmBI recognition sites. When a new LVL0 resistance part is cloned, the chloramphenicol resistance is replaced by the new antibiotic resistance marker resulting in a RFP or sfGFP expressing plasmid with the respective resistance marker. When using these LVL0 parts for LVL1 cloning, the re-ligated resistance parts yield colonies with a visually detectable phenotype. As a result, correct plasmids can be easily identified, even for inefficient LVL1 clonings with < 10 % efficiency.

Choice of fusion sites

Choice of fusion sites for LVL1 cloning

Golden-gate based cloning relies upon the use of type IIs restriction endonucleases like BsaI, BsmBI or BpiI. In comparison to the commonly used type II restriction enzymes (e.g. EcoRI or PstI) they also recognize a specific DNA sequence but cleave outside of their recognition sequence (Pingoud and Jeltsch 2001). The golden-gate-cloning method is taking advantage of this property. A single enzyme can be used to create various single-stranded overhangs that match in a predefined order and finally lead to the correctly assembled plasmid (Engler et al. 2008).
When we started to design the Marburg Collection, we carefully investigated which fusion sites should be used. The fusion sites do not only set the order in which the single parts will be assembled but also affect the assembly efficiency and determine if a newly designed toolbox is compatible with already existing collections, so that parts can be shared easily. For us, the most important and decisive argument was that we wanted to be compatible with as many other toolboxes as possible. It is our strong belief that scientists all over the world should agree on one set of fusion sites to ensure complete interchangeability between different toolboxes. The toolboxes of MoClo (Weber et al. 2011), Loop Assembly (Pollak et al. 2018) and the PhytoBrick system already use a common set of fusion sites. We decided to adapt these fusion sites for all parts that build the transcription unit (Promoter, RBS, CDS, Terminator, Tags). Fusion sites for parts that are novel to our system (Connectors, Oris, Resistance cassettes) had to be newly designed by us because these parts did not exist in the other toolboxes.

We applied the following design principles to obtain optimal fusion sites. Firstly, the newly designed fusion sites must neither be identical to already existing fusion sites nor be palindromic to prevent assembly in a wrong order. Secondly, the fusion sites should not consist of bases that represent a portion of the recognition sequence of a restriction enzyme. If, for example, a fusion site with the sequence GGTC was used, and the sequence of the downstream part starts with TC, a BsaI recognition site would be reconstituted. So all fusion sites that would result in a partial recognition sequence of either BsaI, BsmBI or any of the enzymes that are used in the Biobrick cloning, are excluded. Lastly, the remaining candidates were sorted according to their GC content and the fusion sites with the highest GC content were chosen.
To make design of new parts as simple as possible, we created a collection of overhangs that can be copied from table xxxx and pasted to the sequence specific part of a primer to create new LVL0 parts. These primers contain the cut sites for integration into the part entry vector as well as the predefined fusion sites that are required for correctly assembling LVL1 plasmids. In some cases these overhangs contain additional bases that will be discussed in the following chapter.

View table with primer overhangs

Part Category Fwd Overhang Rev Overhang
1
5’ Connector
AAGGTCTCGCTCGAACACGTCTCGNNNN GGAGTGAGGGAGACCAA
2
Promoter
AACGTCTCGCTCGGGAG TACTTGAGGGAGACGAA
3
RBS
AACGTCTCGCTCGTACTAGAG TAATCAATGTGAGGGAGACGAA
4
CDS
AACGTCTCGCTCGAATG GCTTTGAGGGAGACGAA
5
Terminator
AACGTCTCGCTCGGCTTAA CGCTTGAGGGAGACGAA
6
3’ Connector
AAGGTCTCGCTCGCGCT NNNNGGAGACGAGCTTGAGGGAGACCAA
7
Ori
AACGTCTCGCTCGAGCT TGCTTGAGGGAGACGAA
8
Resistance
AACGTCTCGCTCGTGCTT AACATGAGGGAGACGAA
4x
N-Tag
AACGTCTCGCTCGAATG GGGATGTGAGGGAGACGAA
4y
N-tagged CDS
AACGTCTCGCTCGGATG GCTTTGAGGGAGACGAA
5a
C-Tag
AACGTCTCGCTCGGCTTTA GGGTATGAGGGAGACGAA
5b
Terminator
AACGTCTCGCTCGGGTAA CGCTTGAGGGAGACGAA
Table 1: Primer overhangs for creating new LVL0 parts. Overhangs have to be added to the 5’ end of the primer. Reverse overhangs have to be added as reverse complement. Underlined bases represent BsmBI recognition sites. Bolt bases indicate fusion sites for LVL1 assembly. Bold and underlined bases show BsaI recognition sites. N bases written in bold and italic show fusion sites for lvl1 assembly and have to be custom-designed.

Additional bases between parts

Additional bases between parts

The fusion sites of most golden-gate-based cloning methods create a four base pair scar that is referred to as the fusion site. The fusion sites are the feature that makes toolboxes compatible with each other. Between some parts, additional bases are required for different reasons. These bases were chosen carefully to achieve best performance of the respective part. Please note that these additional bases are not a strict requirement to use or being compatible with our toolbox but we recommend them for the design of additional parts.
The first additional bases were incorporated between the Promoter and RBS part. The fusion site is TACT and AGAG was added additionally. The sequence between promoter and RBS that results by using our suggested overhangs form the same scar that is created if the the parts were assembled with 3A Assembly (Knight 2003). This means that the distance between promoter and RBS is not changed and therefore we do not expect negative effects in transcription or translation. Moreover, we hope that creating the same scar with a different method will make our experimental data more comparable to the data acquired with plasmids assembled with 3A Assembly in previous iGEM projects. The next bases were integrated between RBS and CDS parts. The fusion site, which was adapted from the PhytoBrick system, AATG and TAATC was added upstream of it. Previous work has shown that the sequence between a RBS and the start codon dramatically affects the expression of the desired protein (Lentini et al. 2013). A spacer length of six base pairs was shown to result in the strongest expression. When comparing different bases in a six bp spacer, the experimental data indicate significant differences. The 3A Assembly scar results in 50 % expression strength compared to the sequence TAATCT which was referred to as the reference (Lentini et al. 2013). We chose to use the spacer sequence which is expected to result in highest expression as we think that a system that is designed to enable strongest expression can be easily adapted for low expression by using weak RBS or promoters while going to the opposite direction might be more difficult. Unfortunately, we could not adapt the exact “reference sequence” because the first A in the fusion site AATG is already part of the spacer. Eventually, we used the first five bases of the strongest spacer (Lentini et al. 2013) upstream of the fusion site.
Close attention has to be paid to fusion sites that connect two sequences which are translated like the CDS part or N- and C- terminal Tags. A fusion site in our system consists of four bases which would result in disrupting the triplet code. To prevent mistranslated proteins, two additional bases have to be added to create a six bp scar that results in two translated amino acids. These can be seen as linkers between the joined CDS parts or tags. We decided to preferably use amino acids that are abundant in natural or synthetic flexible linkers like glycine and serine (Chen et al. 2012). Flexible linkers have been shown to improve the performance of epitope tags in Saccharomyces cerevisiae (Sabourin et al. 2007). Therefore we added the bases GG upstream of the fusion site between 4x (N-Tag) and 4y (CDS) which results in glycine and methionine (methionine is preset by the fusion site) and the bases TA downstream of the fusion site between 4 (CDS) and 5a (C-Tag) resulting in an alanine and leucine linker. To allow for the optional use of C-terminal tags, a CDS part must not possess a stop codon. Therefore an additional linker has to be introduced between 4 (CDS) and 5 (Terminator) as well as between 5a (C-Tag) and 5b (Terminator) resulting in an alanine-STOP and glycine-STOP, respectively.
Figure 3: Additional bases and fusion sites ensure correct spacing and allow tags
Between some parts, additional base pairs were integrated to ensure correct spacing and to maintain the triplet code. We expanded our toolbox by providing N- and C- terminal tags by creating novel fusions and splitting the CDS and terminator part, respectively.

Design of Connectors

Design of Connectors

One key feature of the Marburg Collection are the connectors that provide our toolbox with the required flexibility. Our design is inspired by the “Dueber Toolbox”, a golden-gate-based cloning method designed for applications in Saccharomyces cerevisiae (Lee et al. 2015). To our knowledge, the “Dueber Toolbox” is the first cloning system that performs assembly through all levels using restriction sites located on independent basic parts, termed connectors, instead of destination plasmids for level 1 and level 2 plasmids (Weber et al. 2011). Using connectors instead of destination plasmids enables the user to freely choose antibiotic resistance parts and oris. To achieve the flexibility to choose one out of four resistances and one out of three oris, like it is possible in our toolbox, twelve plasmids would be required for building LVL1 plasmids. For building LVL2 plasmids, LVL1 plasmids with different fusion sites are needed. In our system, this is also achieved through the connectors in contrast to the destination plasmids which are used in other cloning methods. In already existing toolboxes, the combinatorics that come with choosing oris and resistances would be multiplied by the number of required positional vectors which provide the fusion sites for LVL2 assembly. A toolbox with five possible positions, four resistances and three oris would require 60 LVL1 destination plasmids. To enable inversion of transcription units, a built-in feature in our toolbox, the previous number would be multiplied by two, resulting in 120 required LVL1 destination plasmids.
Unsurprisingly, this theoretical toolbox would not be convenient to be built and used. The Marburg Collection presents a novel approach that achieves maximum flexibility with a minimum of required components. We provide two sets of connectors. The first set, the “short connectors” solely provide the fusion sites for subsequent assembly. The fusion sites are identical to the ones that are used in building LVL1 plasmids to avoid having to design a complete new set of fusion sites. This also enables using LVL0 ori and resistance parts in LVL2 plasmids. Inversion of individual transcription units was also achieved by designing fusion sites that match in reverse order. The 5’ and 3’ connectors can be chosen independently to build a LVL1 plasmid. This allows the user to combine any number of LVL1 plasmids between one and five into a LVL2 plasmid without end linkers that are required in other toolboxes (Andreou and Nakayama, 2018). In addition to the “short connectors”, we designed a set of “long connectors” that function as genetic insulators. They fulfill the same basic functions as the short connectors, providing fusion sites for subsequent cloning. Additionally, they consist of 300 bp “neutral DNA” flanked by two strong transcriptional terminators. The purpose of these long connectors is to reducing cross interaction between neighbouring transcription units, mainly transcriptional read through, a phenomenon that was described previously (Mairhofer et al. 2015). The design of “neutral DNA” started with generating 300 bp of random DNA which do not possess recognition sites of BsaI, BsmBI and any of the enzymes used in Biobrick Assembly, using a self-made Matlab script. These sequences were analyzed with the reverse mode of the R2oDNA designer (Casini et al. 2014). The R2oDNA designer can take a DNA sequence as input and quantifies the extent of secondary structures, repeats and forbidden sequence motifs, such as promoter or RBS motifs. Because the genome sequence of V. natriegens is not included in the tool, a BLAST search was performed manually with the sequences to check for homologies with the genome of V. natriegens. The top hit was considered as a fourth score to quantify the quality of a spacer sequence. At the end of this process, four scores were assigned to every sequence, each describing a different characteristic. We decided that spacer sequences with decent scores in each category are better suited as insulators than sequences with a superior score in one and inferior scores in other categories.
We developed a Matlab script to find the “best” spacer sequences by increasing quantiles and picking spacers that fall into the lowest quantiles for all four categories. For example, the best sequence was found to be among the best 28 % of all tested sequences in terms of secondary structure, repeats, forbidden motifs and homology to the genome of V. natriegens. The selected sequences were flanked by synthetic transcriptional terminators that were developed in the lab of Christopher A. Voigt (Chen et al. 2013). Most terminators were described to be unidirectional (Chen et al. 2013) and therefore the orientation of the flanking terminators had to be considered. Our priority was to prevent transcription into the respective transcription unit from the upstream sequence. Therefore each spacer sequence was equipped with a strong terminator at the 3’ and 5’ end of the 5’ connector in forward and reverse orientation, respectively.
We expect to achieve the best reduction of crosstalk between neighbouring transcription units by combining our insulators in combination with connectors that facilitate inversion of an individual transcription unit, thus providing a large step towards the reliable and predictable design of synthetic circuits.

Selection of parts

Selection of parts

In addition to the connectors that form the scaffold of our toolbox, we evaluated which parts should be integrated into the Marburg Collection. As a starting point, we focused on parts that are already commonly used in E. coli to demonstrate their compatibility with our new chassis V. natriegens. Therefore we integrated all 20 promoters of the Anderson Library, four RBS from the Community Collection and the most frequently used reporters and terminators. We also included two well established inducible promoters, three widely used oris and the four resistance parts covering the antibiotics used by iGEM.
In addition to this standard set of parts, our toolbox provides N- and C-terminal fluorescence and epitope tags to facilitate fluorescence microscopy and protein purification and degradation tags that can be used to finetune protein levels. Lastly, we submitted all project related parts that can be used to perform genome engineering via CRISPR/Cas9 or natural competence and all enzymes that were used in our metabolic engineering project.

Summary of the features of the Marburg Collection

Summary of the features of the Marburg Collection

The Marburg Collection is designed to be the most flexible golden-gate-based toolbox for prokaryotes. The high degree of flexibility of our toolbox is achieved by the de novo construction of plasmids instead of using entry vectors like most other toolboxes. A novel core feature are the connectors that provide the fusion sites for subsequent cloning steps and can be used to set the order and orientation of each individual transcription unit. A set of newly designed insulators, consisting of “neutral DNA”, and two transcriptional terminators, are designed to minimize crosstalk between neighbouring transcription units.

Metabolic engineering

Our vision of Accelerating Metabolic Engineering

Our Goal

Our goal is to accelerate metabolic engineering by establishing a workflow for fast pathway construction, product screening and pathway optimization. Metabolic engineering is a broad field in synthetic biology, where existing pathways are divided into modular parts and then used and combined to build new pathways ( (Stephanopoulos et al.2012) ). One of the best examples for the impact of metabolic engineering on solving global problems is the engineered production pathway for artemisinic acid, the precursor for a drug against malaria (Ro et al.2006) . In the old days the drug had to be isolated out of the plant Artemisia annua. This was very inefficient because of the limitations of plant growth and low yields of artemisinic acid. Therefore, the group of Jay Keasling engineered a metabolic pathway to produce the drug in yeast cells, which grow faster and produce a lot more artemisinic acid. As in this example, metabolic engineering enables the production of drugs and other valuable products with an increased yield when compared to product isolation from natural sources. Furthermore, the large variety of usable enzymes, pathways and chassis enables more flexibility in pathway design compared to chemical approaches.

Why should we accelerate metabolic engineering?

Although much progress in metabolic engineering was made in the last 20 years, there are still many drawbacks and limitations. For instance, it is crucial to have a big set of genes as combinable parts to find the optimal pathway and they have to be predictable. With many of them just poorly characterized, rational pathway design remains difficult. Our understanding of the working and interactions of enzymes at the level of the metabolism is still rudimentary. Consequently, metabolic engineering consists of much trial-and-error and researchers have to endure series of failures. One strategy to face these issues is modeling of enzyme behavior but currently limitations in computing power prevent this from being used in high-throughput. Another strategy is to use chassis as workhorses to test a lot of enzyme combinations and pathway variants. Here, organisms like E. coli, S. cerevisiae or C. glutamicum are often used but product formation is limited by growth rate and nutrient uptake (Stephanopoulos et al.2007). When a pathway is constructed and to be tested, one has to wait until the production strain has grown and then product formation is still limited by substrate uptake. Another big drawback is that screening for product often is expensive, time consuming and complicated. Conventional methods like LC-MS need a lot of time and resources and it is hardly possible to use it in high-throughput, for instance for a whole pathway library. To overcome these limitations and accelerate metabolic engineering, we established V. natriegens as chassis for metabolic engineering. Its doubling time of under 10 minutes makes it the perfect organism to quickly test many pathway variants or enzyme versions. Additionally, it has a glucose uptake rate of 3.90 g g-1h-1 while E. coli, S. cerevisiae and C. glutamicum have much lower rates (1.90, 3.52 and 0.37 g g-1h-1, respectively). Coupled with the short doubling time, V. natriegens is a powerful chassis not only for research purposes but also for industrial usage, where high productivity is crucial.

Text
Figure 1: Metabolic Workflow established by iGEM Team Marburg 2018

How are we accelerating metabolic engineering?

For optimal execution of metabolic engineering we established a workflow, basing on the DBTL-cycle. We use our Marburg Toolbox to build whole pathway libraries, which can then be transformed into our producer strain VibriXpress. By using biosensors for product screening, we circumvent the expensive and time-intensive usage of LC-MS and learn which pathway variants are the most promising. These findings can then be fed into our pathway-design, thereby closing the DBTL-cycle. To further optimize these pathways, we planned to use directed evolution for more adaptations to that route.

Production of 3-hydroxypropionic acid

3-Hydroxypropionic acid 3HPA

As a proof of concept for our engineering workflow we chose the biological production of 3-Hydroxypropionic acid ( 3HPA ). 3HPA is a compound of high industrial value. In 2004 the U.S. Department of Energy recommended 3HPA as alternative to fossil oils for the r chemical industry (Werpy et al.2004). 3HPA as a platform chemical can be converted into many other compounds like acrylic acid and acrylamide, which both are also precursors for further compounds. According to an estimate of 2014, acrylates alone have an annual market value of USD12 billion, making its precursor 3HPA an optimal target product for our metabolic engineering approach. By finding the optimal pathway to produce 3HPA we could show the efficiency of metabolic engineering in V. natriegens as well as getting a big step closer to an alternative to fossil oils by biological, renewable resources. [Abbildung: T--Marburg--Metabolic_Applications.svg --- noch nicht fertig!]

Our pathway

In recent years many people concentrated on producing 3HPA via many different metabolic pathways (Valdehuesa et al.2013) ; (Vidra et al.2017). Most of these pathways use glycerol or glucose as starting substrate, but there are also publications where acrylic acid, CO2 or uracil were converted to 3HPA (Vidra et al.2017). One promising route starts with glycerol which is dehydrated and oxidized. For E. coli, a productivity of 6.6 mmol g-1 cdw h-1 (oder Lieber andere units? 6.5mmol l-1 oder 0.48 mol mol-1 glycerol) has been shown (Raj et al.2008). However, this route depends on vitamin B12, which often can’t be taken up or produced by the chassis. Many other routes start with glucose, which gets metabolized via propionate, lactate, β-alanine or malonyl-CoA (Vidra et al.2017). We decided to use a pathway, based on the conversion of acetyl-CoA into malonyl-CoA and finally 3HPA . According to (Valdehuesa et al.2013) who evaluated many pyruvate-derived production pathways from a thermodynamic point of view, this is one of the most efficient routes for 3HPA production. This pathway was first established in E. coli, but it was also tried in many other chassis like S. cerevisiae and S. elongatus (Kildegaard et al.2016); (Liu et al.2017); (Rathnasingh et al.2012). We chose this route for several reasons. Firstly, it is based on glucose degradation and V. natriegens has an unbeaten glucose-uptake rate, enabling high glucose consumption. Secondly, vitamin B12 is not necessary, the only cofactors involved are NADPH and biotin, both of which occur naturally in the organism. Thirdly, theoretically just one further enzyme is needed to complete the pathway. All enzymes necessary for the production of malonyl-CoA are already present in V. natriegens and we just have to integrate the last enzyme, malonyl-CoA reductase. Nevertheless, overexpression of acetyl-CoA carboxylase increases production, so we took both enzymes into consideration. (Liu et al.2017) proposed, that there are many ways to further optimize the pathway and direct the flow towards a high product titer. Hence, implementing many variations of this pathway into V. natriegens and testing them in a fast manner promises a wealth of knowledge about pathway optimization and general principles of metabolic engineering.

text
Figure 2: Pathway we have chosen for our metabolic engineering efforts.
https://static.igem.org/mediawiki/2018/thumb/0/0d/T--Marburg--Pathway.svg/1024px-T--Marburg--Pathway.svg.png The first relevant enzyme in the pathway is the acetyl-CoA carboxylase (Acc) that catalyzes the carboxylation of acetyl-CoA. It is involved in fatty acid biosynthesis,consisting of four subunits in E. coli, a biotin-carboxylase (BC), a biotin-carboxyl-carrier-protein domain (BCCP) and two carboxyltransferase subunits (CT) (Jansen et al.2004). During catalysis, a biotin molecule is first linked to a lysine residue in the BCCP domain. Via ATP-hydrolysis the BC domain carboxylates the biotin, which then induces a conformational change and brings the carboxy-group into close proximity of the CT domain. Finally, the CT domain transfers the carboxy-group to the acetyl-CoA, leading to the formation of malonyl-CoA (Lee et al.2008) The subunits are organized in four genes in E. coli (Rathnasingh et al.2012) , however, in other organisms like C. glutamicum the genes are located in two operons (Gande et al.2007).
Text
Figure 3: Mechanism of ACC
The second relevant enzyme in the pathway is the malonyl-CoA reductase (Mcr), which is involved in CO2 fixation by the 3-hydroxypropionate bicycle in thermophilic bacteria (Alber et al.2002). It is a 132kDa large, bifunctional enzyme with alcohol dehydrogenase and aldehyde dehydrogenase function. The C-terminal domain catalyzes the NADPH-dependent reduction of malonyl-CoA into 3-oxopropanoic acid, while the N-terminal domain catalyzes consecutive reduction of 3-oxopropanoic acid to 3-hydroxypropionate (Liu et al.2013).

Biosensors for rapid product screening

One mayor goal of synthetic biology is to make biology engineerable and predictable. Currently metabolic engineering does still involve a lot of trial and error. With the number of possibilities for pathway finetuning this scales very fast, becoming very difficult to tackle with traditional methods.

Also, once we create such a library of metabolic pathway variants, possibly including thousands of strains, we have to find the most promising ones among these. A traditional approach would be to check the production of each strain in its own culture flask. Using for example HPLC or mass spec.

A more suitable, easy way to quickly screen for metabolite concentration in a semi quantitative way are biosensors. One of the easiest to use classes of biosensors are metabolite-responsive transcription factors (Liu et al. 2018). They are usually very specific towards their distinct substrates. In vivo, most of these systems help their native organism to react to environmental clues by regulating expression of for instance catabolic enzymes or importers. In the past, these have ben co-opted by researchers to function as sensors or to control enzyme expression to regulate a pathway depending on the availability of a certain metabolite (Rogers, Taylor, and Church 2016). By using reporter genes like GFP or the LUX operon, we have a measurable output that can be used to select for the best contenders. By running this in several iterations, nature's most powerful optimization tool, evolution, can be applied to further improve the pathway (Williams et al. 2016). Applying fluorescence activated cell sorting (FACS) coupled with V. natriegens incredible doubling time, thousands of generations can be screened for progressively better pathway variants.

So, we decided to put our toolbox full of characterized parts to good use. The Marburg collection contains all the parts needed to find the ideal enzyme balance for highest production. Being able to build a large number of differentially assembled pathways is the answer to creating the most productive strain. Finding the right one in that complexity is another difficulty to overcome, one we solved by finding transcriptional regulators that specifically activate reporter gene expression relative to the concentration of key metabolites, 3-hydroxypropionate and Malonyl-CoA. This gives us a quantifiable output with which we are able to judge the different combinations. Also, utilizing the speed of V. natriegens , the time required for a high throughput screening can be reduced dramatically.

We deemed the concentration of two different metabolites as being critical to our project: 3-Hydroxypropionat, the product itself, and malonyl-CoA, an important intermediate we also plan to overexpress.

Using the information from the paper of Hanko et al. (2017), we settled for the hdpR / hdpH system coming from Pseudomonas putida KT2440. It consists the constitutively expressed HdpR protein (BBa_K2560304). , which is able to bind to the P HdpR and induce expression of a gene. In the native setting, that would be the hdpH gene, encoding for a 3-hydroxypropionate dehydrogenase which enables P. putida to grow on 3HPA . We cloned our reporter gene in its position. This way we got a direct readout of the 3HPA concentrations in the individual cells.

Because our model showed that the low activity of V. natriegens native Acc , malonyl-CoA is likely to be a limiting factor in 3HPA production. Having a direct indication to how much malonyl-CoA is available and how our efforts to introduce heterologous Acc´s succeed in elevating the native levels would be quite helpful. This can be achieved in a quite similar way to the 3HPA measurement. Our literature research quickly gave us several starting points. One of them, a system from Bacillus subtilis , in which the repressor FapR binds the 17bp motif fapO, thereby sterically hindering polymerase binding to an adjacent promoter(Liu et al. 2015). When the FapR protein binds malonyl-CoA, a conformational change is triggered which results in FapR dissociating from the fapO site. The now accessible promoter drives expression of the gene under its control, which we replaced with a reporter.

Directed Evolution and Applications

Directed evolution

Although the Marburg Toolbox enables the construction of many pathway variations, there are still many more possibilities for direct metabolic fluxes towards 3HPA . To enforce these flux improvements, we planned to use directed evolution coupled with FACS selection. By implementing a gene for the enzyme XXX which catalyzes the conversion of 3HPA into succinate, we lead our product back to the central metabolism. By deleting all other genes responsible for succinate production, the cells are forced to use the 3HPA bypass to get succinate, which is needed in the TCA cycle. The big advantage of using V. natriegens for that purpose is that it’s short doubling time increases the number of cell divisions per time and thereby increases the number of mutations in a given time frame. Coupled with our biosensors and fluorescence-activated cell sorting (FACS) we can easily and rapidly select for cells performing the most efficient pathways. Additionally to mutation-guided evolution there are many genome modifications which could increase flow towards 3HPA , for instance by downregulation of side reactions. In the Design-Section a list of possible genome modifications is shown. [Link zu Directed evolution im Designabschnitt machen]. By using MAGE, it is possible to get a library of different genome modifications, which can then be selected via FACS for their efficiency to produce 3HPA or succinate. MAGE stands for multiplex automated genome engineering and is a technique to create combinatorial genome modifications. MAGE uses a set of oligos, which get transformed into a population of cells and then integrate into the chromosome via homologous recombination. Different cells in the population will have different numbers and combinations of modifications leading to a genomic diversity. By applying MAGE in combination with FACS for our pathway we get a huge variety of possible pathway optimizations which can be sorted via FACS based on their efficiency.

Applications of our workflow

By using V. natriegens as a chassis for metabolic engineering, it is possible to build, test and optimize pathways and enzymes at a high-throughput level. The Marburg toolbox enables the fast construction of pathway libraries, while the genetic accessibility of the strain makes genome modification and pathway integration simple. The usage of our biosensors for screening of malonyl-CoA and 3HPA accelerates identification of the best pathway variants and the short doubling time coupled with the high nutrient uptake rate empowers V. natriegens to be the perfect chassis for metabolic and enzyme engineering!

References

Alber, B.E., Fuchs, G., 2002. Propionyl-coenzyme a synthase from Chloroflexus aurantiacus, a key enzyme of the 3-hydroxypropionate cycle for autotrophic CO2 fixation. J. Biol. Chem. 277, 12137–12143.

Gande, R., Dover, L.G., Krumbach, K., Besra, G.S., Sahm, H., Oikawa, T., Eggeling, L., 2007. The two carboxylases of corynebacterium glutamicum essential for fatty acid and mycolic acid synthesis. J. Bacteriol. 189, 5257–5264.

Hanko, E.K.R., Minton, N.P., Malys, N., 2017. Characterisation of a 3-hydroxypropionic acid-inducible system from Pseudomonas putida for orthogonal gene expression control in Escherichia coli and Cupriavidus necator. Sci. Rep. 7, 1724.

Janßen, H.J., Steinbüchel, A., 2014. Fatty acid synthesis in Escherichia coli and its applications towards the production of fatty acid based biofuels. Biotechnol Biofuels 7, 7.

Kildegaard, K.R., Jensen, N.B., Schneider, K., Czarnotta, E., Özdemir, E., Klein, T., Maury, J., Ebert, B.E., Christensen, H.B., Chen, Y., Kim, I.K., Herrgård, M.J., Blank, L.M., Forster, J., Nielsen, J., Borodina, I., 2016. Engineering and systems-level analysis of Saccharomyces cerevisiae for production of 3-hydroxypropionic acid via malonyl-CoA reductase-dependent pathway. Microb. Cell Fact. 15.

Lee, C.-K., Cheong, H.-K., Ryu, K.-S., Lee, J. Il, Lee, W., Jeon, Y.H., Cheong, C., 2008. Biotinoyl domain of human acetyl-CoA carboxylase: Structural insights into the carboxyl transfer mechanism. Proteins Struct. Funct. Bioinforma. 72, 613–624.

Liu, C., Ding, Y., Xian, M., Liu, M., Liu, H., Ma, Q., Zhao, G., 2017. Malonyl-CoA pathway: a promising route for 3-hydroxypropionate biosynthesis. Crit. Rev. Biotechnol. 37, 933–941.

Liu, C., Wang, Q., Xian, M., Ding, Y., Zhao, G., 2013. Dissection of Malonyl-Coenzyme A Reductase of Chloroflexus aurantiacus Results in Enzyme Activity Improvement. PLoS One 8, 1–8.

Raj, S.M., Rathnasingh, C., Jo, J.E., Park, S., 2008. Production of 3-hydroxypropionic acid from glycerol by a novel recombinant Escherichia coli BL21 strain. Process Biochem. 43, 1440–1446.

Rathnasingh, C., Raj, S.M., Lee, Y., Catherine, C., Ashok, S., Park, S., 2012. Production of 3-hydroxypropionic acid via malonyl-CoA pathway using recombinant Escherichia coli strains. J. Biotechnol. 157, 633–640.

Ro, D.K., Paradise, E.M., Quellet, M., Fisher, K.J., Newman, K.L., Ndungu, J.M., Ho, K.A., Eachus, R.A., Ham, T.S., Kirby, J., Chang, M.C.Y., Withers, S.T., Shiba, Y., Sarpong, R., Keasling, J.D., 2006. Production of the antimalarial drug precursor artemisinic acid in engineered yeast. Nature 440, 940–943.

Stephanopoulos, G., 2012. Synthetic biology and metabolic engineering. ACS Synth. Biol. 1, 514–525.

Stephanopoulos, G., 2007. Challenges in engineering microbes for biofuels production. Science 315, 801–4.

Valdehuesa, K.N.G., Liu, H., Nisola, G.M., Chung, W.J., Lee, S.H., Park, S.J., 2013. Recent advances in the metabolic engineering of microorganisms for the production of 3-hydroxypropionic acid as C3 platform chemical. Appl. Microbiol. Biotechnol. 97, 3309–3321.

Vidra, A., Németh, Á., 2017. Bio-based 3-hydroxypropionic Acid : A Review. period. Polytech. Chem. Eng. 1–11. Werpy, T., Petersen, G., 2004. Top Value Added Chemicals from Biomass Volume I — Results of Screening for Potential Candidates from Sugars and Synthesis Gas Top Value Added Chemicals From Biomass Volume I : Results of Screening for Potential Candidates. Other Inf. PBD 1 Aug 2004 Medium: ED; Size: 76 pp. pages.

Biosensors

The design of our biosensors was largely guided by the goal of having as little impact on the host metabolism, while simultaneously being simple and easy to implement. We chose to design all parts to be compatible with the Marburg toolbox, RFC10,RFC25,andRFC1000, to make it as easy as possible for us and future users to implement it and adapt it to new requirements.

Annotation of the location for the regulatory sequences upstream of the HdpR coding sequence was insufficient for us to precisely remove the activator binding site or the constitutive promoter driving HdpR expression. So, we included an about 200bp long region upstream of the HdpR, which should be enough to include all regulatory motifs. Also included is the HdpR coding sequence with its own ribosomal binding site (RBS). and promoter. This results in a part that includes the whole sensing function, but does not quite fit our scheme of clearly defined functional units. Since expression of any part cloned behind the regulatory region is directly controlled by it, and the part serves as both, a promoter and RBS. It is basically an inducible promoter fused to an RBS. Thus, we decided to give the part the overhangs of an RBS. on the downstream and a promoter on the upstream end.

In order to obtain the desired region from the P. putida KT2440 genome, we planned to amplify it via PCR and then isolate the fragment by gel purification of the corresponding band. Into the primer-design we factored in the overhangs we needed to make it Marburg toolbox compatible. So, we gave them non-binding overhangs which automatically added the necessary BsmBI cut sites.

For the malonyl-CoA sensor, our proceeding was very similar. Just that we knew the exact location of all functional parts, except of the location of the promoter, form the 2015 paper by Liu et al and from annotations in the published genome (Liu et al. 2015). Therefore, we could implement it fully into the Marburg toolbox, by simply treating the FapR coding sequence as coding sequence. Not that straightforwardly was the design of the fapO regulator binding site. Since we did not know the promoter strength, and the mechanism of regulation was independent of the promoter sequence, we decided to implement a promoter from the Andersen library, J23100. By placing it right in front of the fapO site, we created a construct that, in theory, should give us a FapR controlled promoter. For the amplification of the individual parts from the B. subtilis 168 genome, again, we used primers sporting non-binding regions, adding the necessary overhangs for our toolbox.

For future projects, other sensors could also be employed. In a publication from 2016, Rogers et al. propose an alternative sensor for 3HP with a lower perception threshold (Rogerset al.2016). We choose not to implement it in our design because of its more complicated mechanism that requires the implementation of more genes. Intriguingly, it works by first converting 3HP to acrylate via a pathway of three heterologous enzymes. The acrylate is then sensed by a transcription factor, similarly to how our sensors described above work. They also report its successful implementation and application in increasing the 3HP titer up to 23-fold. It is conceivable that with a little more effort our system could yield comparable results. Off cause, s we have the additional benefit of having a much shorter iteration interval.

Liu, Di, Yi Xiao, Bradley S. Evans, and Fuzhong Zhang. 2015. Negative Feedback Regulation of Fatty Acid Production Based on a Malonyl-CoA Sensor-Actuator ACS Synthetic Biology 4(2): 132-40. http://www.ncbi.nlm.nih.gov/pubmed/24377365 (October 6, 2018).

Rogers, Jameson K., and George M. Church. 2016. Genetically Encoded Sensors Enable Real-Time Observation of Metabolite Production Proceedings of the National Academy of Sciences 113(9): 2388&-93. http://www.ncbi.nlm.nih.gov/pubmed/26858408 (October 13, 2018).

B. Marchal